Electronics companies are showing a renewed interest in Lean manufacturing principles such as the use of demand-pull replenishment systems, or Kanbans. Although few would argue with the basic premise of waiting until you have demand to pull inventory, some companies are struggling with how to make this work in the extended electronics supply chain. Purists would have Kanbans cascading all the way down to the lowest components, eliminating the need for a regular MRP. This is where the angst begins. Does it make sense to establish a pull system for thousands of lower level electronics components? Should companies abandon their forecasts and MRP when they move to Lean? What are the trade-offs?
At the core of the debate is the generally accepted thinking that forecasts are never right. Of course, the degree to which they are wrong varies depending on the process and the discipline employed to create them. Those who argue for reducing the role of MRP in manufacturing, point to this fundamental inaccuracy and conclude that companies that drive builds based on forecast are probably building the wrong product at any point in time.
The beauty of a pull system is that after the Total Kanban Quantity** is built (or purchased), new builds are triggered in direct relationship to demand. To a certain extent, it is a self-correcting system. If there is no demand for an item, no new supply will be procured. Conversely, if there is increased demand for an item, this increases Kanban pulls and signals an increase in the corresponding supply. The responsive nature of a Kanban can be contrasted against some of the shortcomings of an MRP based approach including:
- Without careful maintenance, an MRP system can trap inventory in "past due" demand. Once the MPS (master production schedule) is loaded, MRP continues to drive to that demand, regardless of how ridiculous, until someone changes it. It is easy for unconsumed demand to continue to march through time until it slips into a window where it is driving needs for material in the past. It is not unusual for us to look at a horizontal MRP and see 2-3 weeks worth of material trapped in a "prior" or "past due" time bucket on the demand line.
- MRP based supply chains are less responsive to product mix changes. To make a change, first you have to recognize that there has been a demand shift - a challenge for even the most seasoned professional, and then you have to change the MPS or the open purchase orders to match. Since most companies have a monthly cycle for forecast and MPS review, the process is not sensitive enough to market changes. What tends to happen is that companies over-load the MPS to capture all potential versions of a product. Once the month or the quarter is over, they see where they have ordered too much and bring down near-term builds to correct inventory. This approach almost guarantees too much inventory and a lot of last minute expediting and re-prioritization.
No debate is complete without a worthy counterpoint. Kanban solutions have their own set of shortcomings. Here are a few of the more serious issues:
- To create the Total Kanban Quantity, the various formulas use some form of average daily demand. This average can be based on historical demand, future demand, or some special mix of the two. The problem is that actual demand does not match the expected average and you end up with too much or too little inventory. This can be especially problematic where you have lumpy demand for a product or where you suffer from some exaggerated demand trend (e.g. a quarter-end demand profile that looks like a hockey stick).
- Maintaining Kanbans takes time and analysis. With MRP, once you set-up the components, the bill of materials (BOM), and load the MPS, you do not need to pay a lot of attention to the lower level components other than ordering per the exploded demand. With Kanbans, you have to manage the calculations to establish Total Kanban Quantity for each product, sub-assembly, and component. Get the calculation wrong and you could be ordering the wrong quantity - this in turn affects inventory and availability. The calculation can become particularly tricky when you have long lead-times, lumpy demand, and parts that are used in multiple products in a high-change environment.
- Kanbans are decoupled. Once you establish Kanbans, they operate independently from one another. Where MRP attempts to bring in matched sets of materials based on the BOM quantities, there is no such connection with Kanbans. As you attempt to manage thousands of lower level components this becomes an issue. In the end, Kanban or demand pull systems are not perfect either. While they have the benefit of responding well to demand, and mix changes, there is the issue of synchronizing the various components and the maintenance of the Kanban quantities that can cause you unexpected headaches.
Some thought-leaders have begun looking to use the strengths of both systems in a sort of hybrid-lean system. In this model, products, sub-assemblies, and critical components are on a pull system. Once a build is triggered for these items, the gating date (based on lead-time or acknowledged delivery) from these items sets an expected build date. This build date is used to drive an MRP demand for dependent, lower-level components creating a matched set. This approach too will have some yet uncovered disadvantages.
To answer the original question - To MRP or Not to MRP - we have to end with a qualified "it depends". Pull systems are more responsive to demand and mix changes. This makes them ideally suited for use at the product and sub- assembly level. MRP based approaches have the advantage of numbers. With today's modern computers it is easy manage thousands of part numbers, the exploded demand, and the resulting actions. The bottom line is that neither approach can claim overall superiority for managing a supply chain. Each has its strengths and weakness: neither can be put on auto-pilot and expected to work.
Lean pull systems
is one of the many issues that Symphony Consulting covers in its workshop
"Reducing
Inventory Liability and Enhancing Flexibility." For more information,
click on the link above or go to Symphony Consulting (www.symphonyconsult.com)